Comment 6 for bug 415508

FWIW, while poolie was reviewing the content filtering changes and assisting with the associated dirstate changes, I'm pretty sure he suggested that path_content_summary() ought to go one day. As best I can tell, there are only two client calls in bzrlib to tree.path_content_summary():

* one in commit.py
* one in transform.py (as it's implementing it's own path_content_summary())

In the second case, there's one call to transform.path_content_summary() and it only uses the kind field in the tuple, not the size or other fields.

So I think deprecating path_content_summary() and the non "record-iter-changes" commit path using it is feasible. Does this need to wait until Robert's selective file commit patch lands first or not?

If there's agreement in principle, I'll check the code bases of the popular plugins for any client calls and ask on the list as to whether there's any objections to it going.